modify psenet scripts

official/cv/psenet/train.py

@@ -118,9 +118,9 @@ def train():
         net = TrainOneStepCell(net, opt)
 
     time_cb = TimeMonitor(data_size=step_size)
-    loss_cb = LossCallBack(per_print_times=10)
+    loss_cb = LossCallBack(per_print_times=step_size)
     # set and apply parameters of check point config.TRAIN_MODEL_SAVE_PATH
-    ckpoint_cf = CheckpointConfig(save_checkpoint_steps=1875, keep_checkpoint_max=3)
+    ckpoint_cf = CheckpointConfig(save_checkpoint_steps=20*step_size, keep_checkpoint_max=10)
     ckpoint_cb = ModelCheckpoint(prefix="PSENet",
                                  config=ckpoint_cf,
                                  directory="{}/ckpt_{}".format(config.TRAIN_MODEL_SAVE_PATH,